How Groups Work In Abstract Algebra

Build a working mental model of groups as symmetry and transformation systems. Learn how axioms show up in real examples, how subgroups and cosets reveal structure, and how homomorphisms, quotients, and actions turn abstract rules into usable tools.

Groups feel abstract until you treat them as machines for doing consistent moves. A group is not just a set with an operation. It is a promise that combining moves stays inside the system, that there is a do-nothing move, and that every move can be undone. The surprise is how little you need for that promise to become powerful. Once you see groups as symmetry machines, subgroups become ways to compress information, and homomorphisms become ways to translate structure without losing what matters.

Groups as symmetry machines

A group is a set $G$ with a binary operation, often written *, that turns any pair of elements into another element of the same set. Thinking in terms of symmetries makes the axioms feel less like rules you memorize and more like constraints that keep the machine from breaking.

The first constraint is closure. If your moves ever produce something that is not an allowed move, you no longer have a self-contained system. The second is that your operation must be one operation, not a different rule depending on which elements you pick. Once you accept that, the rest of the axioms explain what kind of consistency you get when you chain moves.

See how symmetries become elements and composition becomes the operation.

What you should notice is that the elements are not numbers. They are transformations, and the operation is do one transformation after another. That viewpoint is why the axioms are minimal. Each axiom blocks a specific kind of failure you do not want in a calculus of moves.

Moves first
When elements are transformations, closure and inverses stop being abstract. They become the statements you can always perform and undo legal moves.

The group axioms under a microscope

A group $(G,*)$ satisfies four axioms.

• Associativity means $(a*b)*c=a*(b*c)$, so parentheses do not change the result of chaining moves.
• Identity means there is an element $e$ with $e*a=a*e=a$ for all $a$.
• Inverses means each $a$ has an $a^{-1}$ with $a*a^{-1}=a^{-1}*a=e$.
• Closure is already baked into saying * maps $G\times G\to G$.

The common confusions are worth naming precisely.

Associativity is not commutativity

Associativity lets you regroup. It does not let you swap. Commutativity is $a*b=b*a$, and groups that have it are called abelian. Many important groups are not.

Identity is unique, but can look different

In $(\mathbb Z,+)$ the identity is $0$. In a matrix group under multiplication, the identity is $I$. In a permutation group, it is the do-nothing permutation. Same role, different costume.

Inverses depend on the operation

Inverses under addition are negatives. Under multiplication they are reciprocals, when they exist. Under composition they are undo functions. If you change the operation, you changed the meaning of inverse.

Explore how the same operation looks in different representations and where each axiom shows up.

Once you can point to the identity and inverse in each representation, you stop treating the axioms as a checklist and start using them as handles. Associativity is the silent workhorse that makes algebraic manipulation legal.

Canonical examples that generate intuition

Some examples recur because each highlights a different aspect of group behavior.

Four you should be fluent with

• $(\mathbb Z,+)$ is infinite, abelian, and generated by $1$ or by $-1$.
• $(\mathbb Z_n,+)$ is finite, abelian, and cyclic. Its identity is $0\bmod n$.
• $S_n$ under composition is finite and non-abelian for $n\ge 3$. Elements are permutations.
• $GL(n,\mathbb R)$ under matrix multiplication is infinite and usually non-abelian. Elements are invertible matrices.

A useful mental shortcut is to separate the object from the operation. The same underlying set can be a group under one operation and not a group under another. For instance, $\mathbb Z$ is a group under addition but not under multiplication because most integers do not have multiplicative inverses in $\mathbb Z$.

Compare the properties that matter in practice, not just the names.

When you classify an example as finite, cyclic, or abelian, you are predicting what techniques will work. Cyclic groups reduce many questions to arithmetic. Non-abelian groups force you to track order and conjugation behavior instead of only counting.

Subgroups and cosets

A subgroup $H\le G$ is a subset that is itself a group under the same operation. It is a way of isolating a smaller set of moves that is closed under doing and undoing.

Cosets appear when you use $H$ to chunk the whole group into equal-looking pieces. For $g\in G$, the left coset is $gH=\{gh:h\in H\}$. The key point is that cosets are not usually subgroups. They are translates, like shifting a lattice.

Two facts drive most of the structure you extract from cosets.

• Cosets of $H$ either match exactly or are disjoint.
• Every coset has the same size as $H$, so $|G|$ is a multiple of $|H|$ when $G$ is finite.

That second statement is Lagrange’s theorem, and it is less mysterious when you picture cosets as equal-size blocks partitioning $G$.

Explore how choosing a subgroup changes the partition into cosets.

The number of cosets is the index $[G:H]$. Index is a counting invariant that often survives when explicit computation is hard. If you can bound the index, you can sometimes deduce what subgroups are possible at all.

Equal-size blocks
If one coset had a different size, you could not translate $H$ by multiplication without breaking invertibility. Inverses are what force uniform block size.

Homomorphisms, kernels, images, quotients

A homomorphism $\varphi:G\to K$ preserves the operation, meaning $\varphi(ab)=\varphi(a)\varphi(b)$. It is a promise that multiplication or addition in $G$ turns into the corresponding operation in $K$ after you apply $\varphi$.

Two subsets attached to a homomorphism explain almost everything it does.

Kernel and image

The kernel $\ker\varphi=\{g\in G:\varphi(g)=e_K\}$ collects the elements that become invisible under the map. The image $\operatorname{im}\varphi=\{\varphi(g):g\in G\}$ is what you actually hit.

The kernel is always a subgroup, and more strongly it is a normal subgroup, which is the extra condition needed for a quotient group to make sense. Intuitively, normality means the subgroup is compatible with the group’s internal twisting, so cosets multiply consistently.

Isomorphism as perfect translation

An isomorphism is a bijective homomorphism. If $\varphi$ is an isomorphism, $G$ and $K$ are the same group in different notation. You can move questions back and forth without loss.

Simulate how changing a homomorphism changes its kernel and image.

The payoff is the quotient viewpoint. If $\ker\varphi$ captures what gets collapsed, then $G/\ker\varphi$ captures what remains. This is why kernels are not just technical. They measure which distinctions your map refuses to remember.

Group actions and orbit–stabilizer

A group action is a homomorphism from $G$ into a permutation group of some set $X$. Instead of staring at abstract elements, you watch what they do to points in $X$. This is not a different topic. It is the same group, expressed as moves on something you care about.

An action splits $X$ into orbits, the chunks reachable from one another by the group’s moves. For a point $x\in X$, the stabilizer $G_x=\{g\in G:g\cdot x=x\}$ is the subgroup of moves that keep $x$ fixed.

Orbit and stabilizer talk to each other through a counting identity. When $G$ is finite,

$$|G|=|\operatorname{Orb}(x)|\cdot|G_x|.$$

This is the orbit–stabilizer theorem. It turns questions about how many points move into questions about how many symmetries remain once you pin something down.

Visualize an action and see orbits and stabilizers emerge.

Actions are the fastest route from group theory to concrete reasoning. They power counting arguments, symmetry constraints in combinatorics, and geometric classification tricks where the group is hard to list but easy to observe.

How groups fail and what replaces them

Groups are strict, and the strictness is doing work. When an axiom fails, you often still get a useful structure, but with different tools.

A semigroup keeps closure and associativity but drops identity and inverses. A monoid is a semigroup with an identity. These show up when you can compose processes but cannot always undo them, like string concatenation or endomorphisms of a set.

The most common practical failure mode is accidental non-closure. You start with a set of allowable operations, then compose two of them and quietly leave the set. Another is assuming inverses exist just because an operation feels like multiplication. In matrix algebra, non-invertible matrices are exactly where group methods stop and linear-algebra methods take over.

Checklist with purpose
When testing a candidate group, look for inverses first. If they fail, you are usually in a monoid or semigroup, not in a broken group.

Next steps for deeper structure

Once you can move between examples, subgroups, maps, and actions, the next layer is about compressing descriptions and predicting what must exist.

Generators let you describe huge groups from a small set of moves. Presentations turn that into a language of generators and relations, a controlled way to say which combinations collapse. Sylow theorems give existence and counting results for subgroups of prime power order, which becomes a backbone for classification in finite group theory.

A good instinct to develop is to ask what data is invariant under isomorphism. If two groups look different but share the same invariants, try to build an isomorphism. If an invariant disagrees, stop searching and use that disagreement to learn what structure is really present.